Multi-Region / Multi-Account Failover

Same Bedrock model deployed in two regions with automatic failover, or two AWS accounts with different IAM postures for blast-radius isolation.

💡

Tip: Region failover is the first real reliability conversation after a gateway has been live for a quarter. This use case packages it as something you can run, not just diagram.

What this use case covers

  • The same Bedrock model deployed in two AWS regions, fronted by a single LiteLLM alias.
  • Automatic failover when one region degrades — circuit-breaker thresholds, recovery probes, and what counts as "degraded."
  • Two AWS accounts with different IAM postures, for blast-radius isolation between environments or tenants.
  • The trade-off between active-active, active-passive, and regional pinning — and when each one is the right call.

Why it matters

Provider outages are not rare. Region outages are not rare. Account-level IAM mistakes are not rare. A gateway without a failover story is a gateway with one bad day from a public incident.

Coming soon

Implementation coming soon. This use case will ship as a runnable folder in infinite-containers/lite-llm, with a step-by-step lesson here and a companion post and video. Until then, the Bedrock prod profile is the single-region baseline this builds on. Track the full publishing roadmap in USE-CASES.md.